Heat Sink Shootout: heatsinkusa's 3.945 -vs- 4.6

beodrone

beodrone

Active Member
bicit said:
Sanding and polishing seems like a lot more work than simply using a bit of grease. Good sand paper also costs a fair amount.
Click to expand...
Agree'd but longer term when swapping out new LEDS its going to save a lot of grief. That paste is a PITA to put on evenly and to take off when switching out. Once the lights are assembled and ready I wish to avoid having to do that song and dance every time I want to upgrade or change out the color frequencies while testing.. IE 3K vs 5700K etc
 
bicit

bicit

Well-Known Member
I was thinking using a stencil to apply the Tim like they do with smd's. Though it is annoying to clean.
 
beodrone

beodrone

Active Member
Doing a quick google search and finding a lot of thermal 'pads' - are those a good choice to use? I just got off the phone with the chip-lok product manager who seems to think using thermal paste is 'overkill'. Was speaking with him about something unrelated but wanted to throw that tidbit out there.... The chip-lok can remove 5o-10o of heat off the top of the chip according to the product manager (due to it's metal design).
 
Abiqua

Abiqua

Well-Known Member
beodrone said:
Doing a quick google search and finding a lot of thermal 'pads' - are those a good choice to use? I just got off the phone with the chip-lok product manager who seems to think using thermal paste is 'overkill'. Was speaking with him about something unrelated but wanted to throw that tidbit out there.... The chip-lok can remove 5o-10o of heat off the top of the chip according to the product manager (due to it's metal design).
Click to expand...
most have low thermal impedance and that is the problem...the thermal conductivity can be higher # than the common pastes, but the thermal impedance is much lower, meaning it essentially bottlenecks the thermal conductance #.
 
C

churchhaze

Well-Known Member
Abiqua said:
most have low thermal impedance and that is the problem...the thermal conductivity can be higher # than the common pastes, but the thermal impedance is much lower, meaning it essentially bottlenecks the thermal conductance #.
Click to expand...
Thermal impedance is the reciprocal of Thermal conductivity. They mean the same thing. It's like comparing period to frequency. If impedance is higher, conductivity is lower.



 
Abiqua

Abiqua

Well-Known Member
churchhaze said:
Thermal impedance is the reciprocal of Thermal conductivity. They mean the same thing. It's like comparing period to frequency. If impedance is higher, conductivity is lower.



Click to expand...
I get that, but does that mean that pads are calculated by surface area, and when cut down in physical size, isn't the W/m-℃ also reduced?

Reciprocal of the physical area is it not?

Thanks for the diagrams too, they were only showing up when I quoted in my little typee box...so maybe this will work:



:leaf::leaf:server 1: abiqua 0 fuckers :joint:
 
C

churchhaze

Well-Known Member
Conductance is linearly proportional to length, and inversely proportional to cross sectional area. Since impedance is the inverse of conductance, impedance is linearly proportional to the cross section area, and inversely proportional to length.

The same is actually true for electric resistors. Double the width means half the resistance, and double the conductance. Double the length means double the resistance, and half the conductance. This is one of the main reasons why it's been a goal to get mosfet gate lengths smaller and smaller. (at least in the past) The lower parasitic resistance is, the faster the transistors can switch. All gates have a length of the manufacturing process since there's never a good reason to make mosfets in cmos longer than the minimum length.
 
Last edited:
C

churchhaze

Well-Known Member
:wall:

Typo. I must have been sativied as shit when I wrote this. Impedance is proportional to length, and conductance is proportional to cross section area.

I'm pretty sativied right now too, so be careful.

churchhaze said:
Conductance is linearly proportional to length, and inversely proportional to cross sectional area. Since impedance is the inverse of conductance, impedance is linearly proportional to the cross section area, and inversely proportional to length.
Click to expand...
 
Rahz

Rahz

Well-Known Member
Another test was done to determine which of the two heat sinks dissipated heat better passively. For this test I decided to run the 3070 @ 1.05A. I had suspected that the 3.95 would win this round since it seems to be designed for it, however the results were very close. Temps were taken after 1 hour of operation and the probe was then examined to insure contact had been maintained.

4.6 - 54.45C
3.95 - 56.1C

Yes, that's right, the 4.6 performed better than the 3.95 in both tests, marginally in the passive and by a more notable amount in the active. However, the mass of the sinks weren't exact.

4.6 - 1.256 lbs
3.95 - 1.2 lbs

So, perhaps it's best to suggest there is no clear difference between the two when it comes to passive cooling which is a bit of a surprise. I was betting on the 3.95.
 
bicit

bicit

Well-Known Member
Rahz said:
Another test was done to determine which of the two heat sinks dissipated heat better passively. For this test I decided to run the 3070 @ 1.05A. I had suspected that the 3.95 would win this round since it seems to be designed for it, however the results were very close. Temps were taken after 1 hour of operation and the probe was then examined to insure contact had been maintained.

4.6 - 54.45C
3.95 - 56.1C

Yes, that's right, the 4.6 performed better than the 3.95 in both tests, marginally in the passive and by a more notable amount in the active. However, the mass of the sinks weren't exact.

4.6 - 1.256 lbs
3.95 - 1.2 lbs

So, perhaps it's best to suggest there is no clear difference between the two when it comes to passive cooling which is a bit of a surprise. I was betting on the 3.95.
Click to expand...
Thanks for this test, this is very good stuff to know :clap::hug:
 
Rahz

Rahz

Well-Known Member
I have tested one of the ebay 50w radial heat sinks using the same method. These heatsinks are "rated" for 50 watts but after building my first lamp with them I noted the top of the sinks ran a bit warm. They were not hot enough to cause pain but definitely on the far side of warm when cooling Vero 18s @ 1.05A (about 30 watts). In contrast they were notable cooler when I used them with my second lamp running Vero 29s @ .7A. Using one to cool a CXB 3070 @ 1.05 A I noted that they felt just as warm as they do cooling the V18s. I was surprised to find after running the lamp for 1 hour that the test spot was reading 54.45C.

This is surprising for a couple reasons. First, after testing 3 different heat sinks I'm getting about the same temperature on all of them. Second, this radial heat sink is 12 ounces which is significantly lighter than the other two. I'm not sure what to make of it, but I've been stringent enough that I have no good reason to think the test results aren't valid comparatively. Any opinions on this?

I also wanted to test the radial with a fan for the fun of it. The guy doing my metal work has the 100CFM but I happened to have a 130CFM fan here so I wired it up and simply placed it on top of the sink. I tried it blowing up and blowing down and discovered blowing down was slightly cooler at 32C -vs- 33C pulling air through the fins. Because of the air movement around the thermocouple it's not a valid comparison with the previous fan tests, and besides that it's totally impractical. I once modified a computer case with a 350CFM fan on the side directly over the heatsink/fan combo. Oddly it made the core run a few degrees warmer... though there was no mention of this in my entry. :)

http://www.extremetech.com/extreme/84504-crazy-case-mod-contest-meet-the-stragglers

Anyway, the radial heat sink has been sanded down to a smooth surface so tomorrow I will remove the thermal compound and do a passive test without it per beodrone's request. I will also be trying the test with a .7A driver as I think there is at least one member here who would find that particularly interesting... and it will give a good counter point to the similar readings I've gotten @ 1.05A.

Also, friendly tip: Don't like using oil to lubricate your drill points? Try some propylene glycol (ejuice). Works fine and rinses off with just water.

radialheattest.png radialwfanheattest.jpg
 
alesh

alesh

Well-Known Member
Rahz said:
I have tested one of the ebay 50w radial heat sinks using the same method. These heatsinks are "rated" for 50 watts but after building my first lamp with them I noted the top of the sinks ran a bit warm. They were not hot enough to cause pain but definitely on the far side of warm when cooling Vero 18s @ 1.05A (about 30 watts). In contrast they were notable cooler when I used them with my second lamp running Vero 29s @ .7A. Using one to cool a CXB 3070 @ 1.05 A I noted that they felt just as warm as they do cooling the V18s. I was surprised to find after running the lamp for 1 hour that the test spot was reading 54.45C.

This is surprising for a couple reasons. First, after testing 3 different heat sinks I'm getting about the same temperature on all of them. Second, this radial heat sink is 12 ounces which is significantly lighter than the other two. I'm not sure what to make of it, but I've been stringent enough that I have no good reason to think the test results aren't valid comparatively. Any opinions on this?

I also wanted to test the radial with a fan for the fun of it. The guy doing my metal work has the 100CFM but I happened to have a 130CFM fan here so I wired it up and simply placed it on top of the sink. I tried it blowing up and blowing down and discovered blowing down was slightly cooler at 32C -vs- 33C pulling air through the fins. Because of the air movement around the thermocouple it's not a valid comparison with the previous fan tests, and besides that it's totally impractical. I once modified a computer case with a 350CFM fan on the side directly over the heatsink/fan combo. Oddly it made the core run a few degrees warmer... though there was no mention of this in my entry. :)

http://www.extremetech.com/extreme/84504-crazy-case-mod-contest-meet-the-stragglers

Anyway, the radial heat sink has been sanded down to a smooth surface so tomorrow I will remove the thermal compound and do a passive test without it per beodrone's request. I will also be trying the test with a .7A driver as I think there is at least one member here who would find that particularly interesting... and it will give a good counter point to the similar readings I've gotten @ 1.05A.

Also, friendly tip: Don't like using oil to lubricate your drill points? Try some propylene glycol (ejuice). Works fine and rinses off with just water.

View attachment 3391987 View attachment 3391988
Click to expand...
Radial HS is way more suited for passive cooling. It's just that you chose three very similar performing (in passive mode) HSs. Nice test, thanks.

I wouldn't like to the temperature on that heat sink with generic COB run at 50W.
 
beppe75

beppe75

Well-Known Member
do you guys think is ok to use the pre applied mx-4 paste that comes with the alpine 64 plus coolers?
or is it better to clean that up and use some other thermal paste...?
 
You must log in or register to reply here.
Top